Hyperfocal distance, focusing, etc.

[brycimus]

So lately I've been practicing at getting everything in an image tack sharp, but I can't seem to succeed. I know that the consensus is that f/8 is the magic spot for sharpness. I've also been doing reading and thread searching about hyperfocal distance. I understand that it's the point where (with a given aperture and focal length) everything in the image from half the HFD to infinity will be in focus. But I don't understand how this is useful when many dslr lenses don't have distance scales on them. I'm using a Sigma 17-70 and the only scale it gives me is .65-1-1.5-3-7------infitiy (ft.) how can I choose the HFD? Also how does autofocus play into the HFD? I know I should focus at a specific place, if say i'm doing a landscape, and I wouldn't know where to focus for everything to turn out well, I'd probably end using the middle focus point.

Thanks

[René Damkot]

So lately I've been practicing at getting everything in an image tack sharp, but I can't seem to succeed. I know that the consensus is that f/8 is the magic spot for sharpness.

It might be that above f/8 the lens gets a bit less sharp, or diffraction comes into play, so *maximum* sharpness is reduced, but at f/11 or f/16 the Depth of Field (DoF) will be greater. So an object that's just outside DoF, will be sharper when you stop down further.

Life's full of compromises ;)

I've also been doing reading and thread searching about hyperfocal distance. I understand that it's the point where (with a given aperture and focal length) everything in the image from half the HFD to infinity will be in focus.

Not "in focus"; within DoF. You could say: Unsharp, but acceptably so.

But I don't understand how this is useful when many dslr lenses don't have distance scales on them. I'm using a Sigma 17-70 and the only scale it gives me is .65-1-1.5-3-7------infitiy (ft.) how can I choose the HFD?
Simple answer: You can't.
You guess and hope for the best.

Also how does autofocus play into the HFD? I know I should focus at a specific place, if say i'm doing a landscape, and I wouldn't know where to focus for everything to turn out well, I'd probably end using the middle focus point.
Again: guess and hope.
Focus somewhere about 1/3 to half into the scene.

Alternatively (easier, and not worse IMO), (auto)focus on infinity (or furthest subject), and let DoF take care of the foreground. The subject matter in the background is smaller on the sensor, so has finer details that need to be resolved, and will appear OoF faster then the (bigger) subjectmatter in the foreground...

Some threads: One (http://photography-on-the.net/forum/showthread.php?t=418104), two (http://photography-on-the.net/forum/showthread.php?t=420772), three (http://photography-on-the.net/forum/showthread.php?t=512771).

Some more reading: Scroll down, and look in the lower left corner "similar threads"

[brycimus]

you say 1/3 - 1/2 into the scene, should i use the af point that matches with this approximation, or do you manual focus?

[René Damkot]

Whatever works for you; pick an AF point, focus, then recompose, or use MF.
It's not an exact science.

Check the links; there are about as many opinions as there are options ;)

[Doug Pardee]

Hyperfocal focusing for landscapes has been pretty well discredited, even for film. For digital it has no theoretical basis at all, because you can't calculate the final sharpness of the digital image so the notion of "acceptably unsharp" cannot be quantified.

The best and simplest approach is to focus on the farthest object(s) that you want to have come out sharp. Stop down as far as you're comfortable, and there you go.

[Lowner]

Doug,

Can you point me in the direction of proof. I am a firm believer in the concept that lies behind the DoF figures, so will need a lot of convincing.

100% of my output is printed, so whether I use film or digital methods to get to that point is surely immaterial. And the "Circle of Confusion" remain as valid as it has ever been.

Richard

[René Damkot]

Richard, have you read the 3 threads I linked in post #2?

[Lowner]

Rene,

I hadn't, but have now.

Paraphrasing and misquoting is not proof. I'm predisposed to mistrust anyone trying to sell me anything, and it feels as though I'm being sold a pup.

I see nothing wrong with the concept that the brain can be "fooled" into thinking that something below a certain size seen at a certain distance is sharp, even when it is not. All I want is for someone who claims that it is not true to show me the proof.

And your links (thanks for them by the way) show a lot of discussion but no hard proof.

I'll accept that it's a moving target. Ansell Adams work with the best lenses of his day produced work that these days seems slightly "off", but that does not and never will proof the CoC idea wrong.

Richard

[René Damkot]

Agree about the "lot of discussion" bit ;)

No hard proof, but from the times I've tried, in my eyes the loss fine detail in the distance was more important then the small win in sharpness the foreground when I (tried to) use hyperfocal distance.

Off course, I don't know how that would translate to a print, since this was shooting tethered to a laptop, and looking at 100% ;)
If I ever shoot a landscape, I'll give both a try...

So, for print, hyperfocal might work, but when pixel peeping on a screen, I tend to follow Merklinger's advice, and focus on the farthest detail.

[nwa2]

Hyperfocal distance focusing works as well with digital as it does with film, why? - because the light focusing end of the camera works in the same way.

I find the technique useful for low light situations where the autofocus system has trouble working and manual focus through the small viewfinders on APC cameras is at best problematic. My 40d is better in low light but my previous 350d was appauling.

I have created a highperfocal chart for all my zooms. I judge my point of focus by eye and set the lens focus using my chart.

[Wilt]

I have created a highperfocal chart for all my zooms. I judge my point of focus by eye and set the lens focus using my chart.

But the OP wonders how you possible set to HFD when the distance scale on the modern AF lens is hardly usable? For example, 50mm f/8 has HFD of 65', and on my 17-55mm f/2.8 lens, 5 feet is the last distance before Infinity -- so where is 65' on the scale!
Even a fixed FL lens like my 100mm has 15' as the longest distance, and a HFD at f/11 of 165'

While it might be possible to guess a distance to focus on, while on land, sometimes we do not have that crutch...imagine focusing at 165' point in the middle of the Grand Canyon or in the middle of a featureless body of water!

[brycimus]

you could always get a tape measure for those landscape scenes :)

so it sounds like focusing on the most distant object and stopping down is the best option

[yogestee]

Hyperfocal focusing for landscapes has been pretty well discredited, even for film. For digital it has no theoretical basis at all, because you can't calculate the final sharpness of the digital image so the notion of "acceptably unsharp" cannot be quantified.

The best and simplest approach is to focus on the farthest object(s) that you want to have come out sharp. Stop down as far as you're comfortable, and there you go.

Show us where its been discredited?? Is there a link somewhere??

[20droger]

The hyperfocal shooting argument is unresolvable. Some very good photographers swear by it, others, equally good, swear at it.

You should learn how and why it works, experiment, and then do what is best for you. If it works for you, great! Use it. If it doesn't work for you, by all means abandone the idea.

It works for us, so we use it.

To more fully understand the concepts, logic, and math behind hyperfocal distance and depth of field in general, I recommend reading this (http://doug.kerr.home.att.net/pumpkin/Depth_of_Field.pdf) paper.

[E-K]

Hyperfocal focusing for landscapes has been pretty well discredited, even for film. For digital it has no theoretical basis at all, because you can't calculate the final sharpness of the digital image so the notion of "acceptably unsharp" cannot be quantified.

The CoC can be quantified however. You pick a CoC you want for a given maximum print size and sensor size combination. The math isn't wrong. Applying things blindly (like using the DoF markers on a lens on an APS-C sensor for a 16x20 print) is wrong.

The best and simplest approach is to focus on the farthest object(s) that you want to have come out sharp. Stop down as far as you're comfortable, and there you go.

Simplest, yes, but now always the best.

e-k

[Lowner]

A few posts back the question was asked "so where is 65' on the scale!", assuming we would all understand how impossible it is to find the hyperfocal distance.

But to my mind, judging 65 ft by eye is no big deal, just pick out an object that far away and focus on that. It does not even need to be in the eventual shot.

Richard

[Wilt]

A few posts back the question was asked "so where is 65' on the scale!", assuming we would all understand how impossible it is to find the hyperfocal distance.

But to my mind, judging 65 ft by eye is no big deal, just pick out an object that far away and focus on that. It does not even need to be in the eventual shot.

Richard

Judging by the large number of people tailgating on the highway with one or maybe two car length's distance to the car in front, travelling at 75-85mph, they are distance impaired as well as time impaired, and could not get the distance right! (half a car length per 10 mph, or two seconds elapsed time)

[Wilt]

Show us where its been discredited?? Is there a link somewhere??

Not to defend the contention, but merely to explain what might be behind that claim...

Hyperfocal distance concept suffers the flaw that it depends (simply) upon COC size in the calculation. We must all remember that the concept of 'maximum acceptable COC' depends upon the individual's visual acuity, and that the application of 'maximum acceptable' will put things at infinity to be on the hairy edge of 'sharp' (unsharp to some) even though in landscape shots the expectation is that things at infinity are truly sharp!

[Glenn NK]

While we can endlessly debate the merits of hyperfocal distance and the usefullness of the DOF calculators and charts, there is something that can be learned from them by playing around with various f/stops and focal lengths:

Focal length has a huge effect on DOF.

Could this be the source of the time-honoured tradition of using wide angle lenses for landscape work?;)

On previous threads on this topic (referenced by Rene Damkot), I have quoted from a book I have by Tim Fitzharris - he seems to have it right because I keep seeing his landscape images on calendars and they all have superb DOF.

[Wilt]

I just realized another factor that might contribute to the 'myth' about hyperfocal distance...

Back in the days of fixed FL lenses and DOF scales engraved next to the distance scale, the common wisdom was to use the DOF scale marks for an aperture ONE WIDER than the set aperture. Yet the DOF and hyperfocal calculations are calculated for the set aperture and do not provide a margin of error which reduces the 'blurriness' of the COC to a smaller size. It is no wonder that someone today would proclaim a 'myth' status to hyperfocal distance calculation.

[yogestee]

I just realized another factor that might contribute to the 'myth' about hyperfocal distance...

Back in the days of fixed FL lenses and DOF scales engraved next to the distance scale, the common wisdom was to use the DOF scale marks for an aperture ONE WIDER than the set aperture. Yet the DOF and hyperfocal calculations are calculated for the set aperture and do not provide a margin of error which reduces the 'blurriness' of the COC to a smaller size. It is no wonder that someone today would proclaim a 'myth' status to hyperfocal distance calculation.

Ok Wilt I agree with this one.. I came up through where lenses both fixed focal length and zooms had distance scales on them relavant to apertures.. It was easy to focus, look down at the scale and determine the depth of field for the aperture I was using.. Also we must bear in mind there is a difference between acceptable sharpness and depth of field..

Using these lenses when focused on infinity and using a realitively small aperture the DOF (acceptable sharpness ?) stretches beyond infinity..

Hyperfocal distance is not a myth.. Its just the current crop of digital lens manuafacturers have omitted the distance scale on their lenses.. Many photographers who came up via the digital age aren't aware these scales ever existed..

On a similar point,, the acceptable sharpness (DOF if you like) stretches farther beyond the point of focus than it does in front of the point of focus.. Hardened landcape photographers use this principle when shooting.. Most will focus at around 1/3rd into the scene and not at infinity knowing especially when shooting at smaller apertures that from this point to infinity will be acceptably sharp..

[Wilt]

On a similar point,, the acceptable sharpness (DOF if you like) stretches farther beyond the point of focus than it does in front of the point of focus.. Hardened landcape photographers use this principle when shooting.. Most will focus at around 1/3rd into the scene and not at infinity knowing especially when shooting at smaller apertures that from this point to infinity will be acceptably sharp..

And unfortunately it is very easy to show that situations where the DOF is 10% in front and 90% behind the point of focus, which makes the 33:66 fule of thumb DOF split a total fallacy!

For one example, 30mm lens at f/8 on APS-C...HFD is at 23', the DOF zone extends from 12'-763', or 1.5% in front of HFD and 98.5% behind!

[yogestee]

And unfortunately it is very easy to show that situations where the DOF is 10% in front and 90% behind the point of focus, which makes the 33:66 fule of thumb DOF split a total fallacy!

Wilt,,I suppose it also depends on the subject and how far you are away from the main point of interest..

[Doug Pardee]

The CoC can be quantified however. You pick a CoC you want for a given maximum print size and sensor size combination. The math isn't wrong.

The problem is that if you pick a CoC that is appropriate, it's so small that it gives you almost no effective DoF.

The conventional figure for CoC gives a resolution at infinity that is less than that of a 640x480 cellphone camera. A more appropriate CoC for a 10-megapixel APS-C sensor would be around .002 mm. With a CoC that small, there is precious little room for focusing error. It is critical to focus on whatever needs to be sharp, which in landscape photos is almost always whatever is in the distance.

Another reason to focus on the farthest item of interest is that the blurring behind the focus point increases rapidly with distance, while blurring in front of the focus point increases very slowly with distance. Any error in focusing should be made toward farther away.

As Merklinger showed, it is the low amount of near-field blurring that makes high DoF landscape photos possible. The best technique is to focus on the most distant object of interest, stop down as far as you are comfortable diffraction-wise, and let the near-field objects sort themselves out. Merklinger demonstrated that objects of the size of the entrance pupil (focal length divided by f-number) would be resolved regardless of distance as long as they were closer than the focus distance—we're ignoring diffraction effects, of course. For a 22mm lens at f/11, objects of 2mm or larger will be resolved no matter how close they are.

[Doug Pardee]

Hyperfocal distance focusing works as well with digital as it does with film, why? - because the light focusing end of the camera works in the same way.

However, you aren't interested in the light focusing per se. You're interested in the image that you capture, not the image that's projected on the sensor. It's not possible with digital to calculate how sharp the resulting image will be based on the sharpness of the projected image. There are too many unknown variables: the anti-aliasing filters, the quantization effect of the digital sensor array (this one is known but a bit difficult to deal with), the demosaicing process, and the real killer is the various steps of sharpening that digital images are subjected to.

In addition, film photos were rarely printed above 8x10, and the hyperfocal focusing technique was designed around contact prints (unenlarged prints, for those unfamiliar with film). Digital photos are often cropped and often printed much larger than 8x10, requiring considerably sharper captures than with film.

[yogestee]

In addition, film photos were rarely printed above 8x10, and the hyperfocal focusing technique was designed around contact prints (unenlarged prints, for those unfamiliar with film). Digital photos are often cropped and often printed much larger than 8x10, requiring considerably sharper captures than with film.

I disagree with you Doug on this point.. How about I come over to your house and show you my film portfolio of 16"x20" prints?? All my collegues exhibted 16"x12" or 16"x 24" prints both colour and B/W..

[E-K]

The problem is that if you pick a CoC that is appropriate, it's so small that it gives you almost no effective DoF.

The conventional figure for CoC gives a resolution at infinity that is less than that of a 640x480 cellphone camera. A more appropriate CoC for a 10-megapixel APS-C sensor would be around .002 mm. With a CoC that small, there is precious little room for focusing error.

The resolution of the 40D is about 0.0057mm. So you are saying I should select a CoC that is smaller than the possible resolution of the sensor? And that's not even taking the effect of the anti-alias filter into consideration.

It is critical to focus on whatever needs to be sharp, which in landscape photos is almost always whatever is in the distance.As I've said each time, there are instances where one method is a better choice than another. There are definitely cases where you need things acceptably sharp close up as well as far away. What you can get away with depends largely on the subject. Fine detail in the foreground which appears blurry may in some cases be unacceptable.

Another reason to focus on the farthest item of interest is that the blurring behind the focus point increases rapidly with distance, while blurring in front of the focus point increases very slowly with distance. Any error in focusing should be made toward farther away.Don't disagree with the erring on the far side.

As Merklinger showed, it is the low amount of near-field blurring that makes high DoF landscape photos possible. The best technique is to focus on the most distant object of interest, stop down as far as you are comfortable diffraction-wise, and let the near-field objects sort themselves out. Merklinger demonstrated that objects of the size of the entrance pupil (focal length divided by f-number) would be resolved regardless of distance as long as they were closer than the focus distance—we're ignoring diffraction effects, of course. For a 22mm lens at f/11, objects of 2mm or larger will be resolved no matter how close they are.Yes, but a 2mm resolution can be somewhat disconcerting on some things close-up and I wouldn't expect to be able to resolve something to 2mm at infinity.


e-k

[rdenney]

The problem is that if you pick a CoC that is appropriate, it's so small that it gives you almost no effective DoF.

Throughout this thread, estimates for acceptable circles of confusion have been taken as an assumption. They are not. They are determined according to the anticipated ultimate size of the print and the expectation of the photographer.

It's possible for the circle of confusion to be hashed by the pixel array on a given sensor, in which case enlargeability will be controlled by the pixel density rather than by the proximity to sharp focus by the lens. In those cases, digital overrides calculations based on circles of confusions, just as it did in the old days using grainy film, especially with grain-softening developers (e.g., Tri-X in smaller formats developed in Microdol-X, which was popular in the old days, but not for making big prints).

Another problem with digital is that we observe and evaluate sharpness quality not based on print standards, but rather on monitor standards. In other words, we expect stuff to be sharp at 100% on a 100-dpi monitor. For many cameras, that's a huge enlargement. Even for my 10D, that's a 20x30" print, far larger than anyone would attempt to put on paper from an APS-C sensor, no matter how many pixels it had, unless they could be sure the print would not be viewed closer than from across the room.

And then there are the standards for viewing prints. Do we adopt a sharpness standard for making a print seem sharp at proper viewing distance? Or do we set the standard for minimum viewing distance? For the latter, the print must be able to resolve 5 lines/mm to seem sharp to a person with good vision inspecting the print closely. If your camera can resolve 1900 line pairs/APS-C frame, that frame can therefore support 126 line pairs/mm, or a 25x enlargement (which is only 15x23 inches in the print). Few lenses can keep up with that degree of enlargement. On my 5D, however, that 1900 line pairs per frame is only 79 line pairs/mm, which is well within good lenses at sharp focus. For a 5 line/mm print sharpness, the 5D supports a 16x enlargement, which makes--surprise--the same size print as above. But now the lens faults won't be a problem. Being out of focus, however, will.

That's why we get lost with discussions centered on circles of confusion. They are always there, but they are not always relevant. We have to consider the whole system as a system, and in some situations, we are constrained by the sensor, and in others by the lens.

Back to focus. Remember that the lens is at its sharpest only in the plane of sharp focus. Stopping down does not bring things into focus. It ONLY brings things to an acceptable level of apparent focus. Only those items in the plane of sharp focus reach the potential sharpness of the lens. Everything else is a management issue.

I rarely use hyperfocal focusing with modern cameras, because there is always a subject that I want to be sharper than the rest, even if I want it all to appear sharp. I will frequently consider depth of field, stopping down as necessary, but I still make sure to focus on the important part of the subject. I will then chimp as necessary to check the rest. After all, I can't put a 10X loupe on the ground glass like I used to do with view cameras. If I can't get enough depth of field without using such as small aperture that I get some diffraction, I just know that I can't print that image as large. Getting it in apparent focus is a dynamic balance with diffraction, but at least diffraction affects the whole image evenly. So, if I have diffraction, printing small enough to hide it won't undermine the image the way having an fuzzy foreground will.

In the end, I get what I can from the camera and let the results dictate the potential size of the print. That means I can't make my final judgment based on what I see at 100% on a computer monitor.

I still use hyperfocal focusing on some cameras, though. For example, when I use my Nikonos, which has no focusing aid, it's about the only choice for general photography. But we are not so constrained with digital cameras.

Rick "who expects full sharpness only where the lens is sharply focused" Denney

[René Damkot]

Summons it up quite nicely I'd say :)

[Wilt]

Summons it up quite nicely I'd say :)

'Sums' it up, René. No doubt your English is many orders of magnitude better than my Dutch, so I shouldn't be correcting you! ;)

[René Damkot]

Hehehe.
Seems I do learn something here everyday ;)

Thnx.

P.S. and totally OT: What does "summon" stand for? Or is that what you do with ghosts / demons / spirits?

[20droger]

"Summon" is a synonym for "call upon" or "call," as in "the court summoned the defendant," or "he summoned all his courage."

A "sum," on the other hand is the total of numbers, items, or actions. Thus the phrase "that sums it up" indicates that all the listed items together resulted in the (complete) answer.

And, à propos de rien, I, too, am a René.

[René Damkot]

Ah, okay.
Thanks René :)

[20droger]

And I guarantee that your English is far, far superior to my Dutch!

[Wilt]

Hehehe.
Seems I do learn something here everyday ;)

Thnx.

P.S. and totally OT: What does "summon" stand for? Or is that what you do with ghosts / demons / spirits?


OK, René...time for equal opportunity for 20droger and myself! Please provide the Dutch word for 'sums' (verb) and for 'summons' (noun) and we will consider that we have gotten fair compensation for our English lessons to you so that we don't have to send an invoice for our services ;)

[René Damkot]

Letme see: "That sums it up nicely" could be translated to "Dat is mooi samengevat"

("samenvatten" = "to summarise")

Summons: A bit less straight forward.

"the court summoned the defendant" probably would be something like "De rechtbank riep de verdachte op". Where "verdachte" = "suspect" literally, and "oproepen" can also be "to call".
"he summoned all his courage" = "Hij verzamelde al zijn moed"

(Moed = courage, but "verzamelen" = "to collect")

The big challenge would be for you guys to actually speak those sentences :)

[rdenney]

The big challenge would be for you guys to actually speak those sentences :)

No way. Too blasted many consonants; the opposite problem with French, that has too blasted many vowels.

Rick "who can swear and order food in Spanish but that's about all" Denney

[Wilt]

No way. Too blasted many consonants; the opposite problem with French, that has too blasted many vowels.

Rick "who can swear and order food in Spanish but that's about all" Denney

Funny, I have always found Dutch to have both too many vowels (e.g. 'mooi' ) and too many consonents (e.g. 'zijn'). I have also found Netherlands perplexing in that there are lots of photo opportunities showing evidence of dogs (their droppings all over the sidewalk) but never any photo opportunities of the dogs themselves! But yes, if God had intended for me to pronounce all of those words, I would have been raised wearing wooden shoes (which are designed to walk thru the dog droppings, I imagine!)

[René Damkot]

(which are designed to walk thru the dog droppings, I imagine!)

Nah, they were designed to float I'd say. Probably has to do with the weather we're having right here at the moment ;)

I think people would give rather strange looks if I went into town wearing wooden shoes... I do wear them in the garden sometimes :lol:

They are rather old fashioned / touristic; most people in towns don't use them any more, and not too much people in the countryside either. (Which kinda brings us back on topic: DoF in landscape shots ;))

[SlowBlink]

Thanks Rick, that was the only post I could fully understand.

note: I can say "Your dog smells like cheese" in Italian and German but I can barely speak my own until I've had a couple cups of coffee. :)

[20droger]

Ah! But those wooden shoes are great for screwing up machinery! (Look up the origins of the word "sabotage"!)

[zilch0md]

Hi Brycimus,


I'm using a Sigma 17-70 and the only scale it gives me is .65-1-1.5-3-7------infitiy (ft.) how can I choose the HFD? Also how does autofocus play into the HFD? I know I should focus at a specific place, if say i'm doing a landscape, and I wouldn't know where to focus for everything to turn out well, I'd probably end using the middle focus point.
Thanks

I'd like to answer your questions by referring to some posts I made in a different thread. These methods are not everyone's cup of tea and certainly not the way to go for any photography that requires spontaneity, but given that you've expressed an interest in "tack sharp" landscape photography, you might find this appealing.

http://photography-on-the.net/forum/...4&postcount=25

http://photography-on-the.net/forum/...3&postcount=39

http://photography-on-the.net/forum/...5&postcount=45

http://photography-on-the.net/forum/...3&postcount=54

http://photography-on-the.net/forum/...3&postcount=57

Mike Davis

[440roadrunner]

Your posted links don't work

[zilch0md]

Your posted links don't work

Bummer. Thanks for bringing that to my attention.

These will work:

http://photography-on-the.net/forum/showpost.php?p=2756604&postcount=25

http://photography-on-the.net/forum/showpost.php?p=2780643&postcount=39

http://photography-on-the.net/forum/showpost.php?p=2784205&postcount=45

http://photography-on-the.net/forum/showpost.php?p=2786373&postcount=54

http://photography-on-the.net/forum/showpost.php?p=2788383&postcount=57

Mike Davis